Optical disks are produced by making a master which has a desired surface relief pattern formed therein. The surface relief pattern is created using an exposure step (e.g., by laser recording) and a subsequent development step. The master is used to make a stamper, which in turn is used to stamp out replicas in the form of optical master substrates. As such, the surface relief pattern, information and precision of a single master can be transferred into many inexpensive replica optical disk substrates.
During the mastering exposure step, the mastering system synchronizes the translation position of a finely focused optical spot with the rotation of the master substrate to describe a generally concentric or spiral pattern of a desired track spacing or “track pitch” on the disk. The generally spiral track forming the desired surface relief pattern as a result of the mastering process can be defined by high regions termed “lands” and lower adjacent regions termed “grooves” and/or pits (i.e., interrupted grooves). The recording power and size/shape of the focused optical spot (spot size) as well as the photosensitive material parameters determine the final geometry revealed in the master disk during the subsequent development step. Normal mastering practice uses high contrast positive photoresist for the photosensitive material.
Conventional mastering typically utilizes laser light with wavelength, λ, in a range of 350 nm<λ<460 nm focused through an objective with a numerical aperture (NA) of 0.75 nm<NA<0.90 to give a theoretical Gaussian spot size of:                SS=0.57λ/NA (full width at half maximum intensity (FWHM)).Thus, a 350 nm laser light with NA=0.9 gives a theoretical spot size 0.22 microns (FWHM) as the practical limit for conventional optics.        
After the master is recorded, it is flooded with developer solution to reveal the exposure pattern applied by the master recording system. The dissolution of the photoresist in the developer solution is in proportion to the optical exposure previously received in the recording process. The dissolution rate of the photoresist can be modeled for given exposure and development conditions (see Trefonus, P., Daniels, B., “New Principal For Imaging Enhancement In Single Layer Positive Photoresist,” Proc. of SPIE vol. 771 p. 194 (1987), see also Dill F. et al., “Characterization of Positive Photoresists” IEEE Transactions on Electronic Devices, vol. ED-22 p. 445 (1975).) Expressions explained in these referenced technical papers can be used to model the effects of exposures from several adjacent tracks recorded in the photoresist and subsequently developed. The photoresist dissolution in the developer solution is in proportion to the optical exposure previously received (positive type resist). More accurately, the dissolution rate (R) is given by the Trefonas model asR [nm/sec]=R0×(1−M)q+Rb 
Where R0 and Rb are the dissolution rates of the fully exposed and unexposed photoresist (respectively), q is a resist parameter related to the resist contrast and M is the fractional unconverted photoactive compound in the resist. Typical values for commercially available resists are q=3, 10<R0<200 [nm/sec] and Rb=0 for normal developer concentrations. The M term is dependent in a point-wise fashion on how much exposure was received in the resist (E(x,y,z)) and the resist's parametric sensitivity “C” per the Dill convention:M(x,y,z)=exp{−C×E(x,y,z)}.
Since optical disk mastering typically uses only 50-200 nm of photoresist thickness, the-z-dependence of exposure can safely be ignored so that the above equations can be combined to giveR=R0(1−exp{−CE(x,y})q;or, with the exposure profile explicitly circular gaussian we may simplify toR=R0(1−exp{−CkPexp[−r2/SS2]})q;
Where r measures the radial distance from the center of the spot (r2=x2+y2), P is the recording power and k is a normalization constant for the guassian function. This dissolution rate, multiplied by the development time (td), gives the depth of photoresist lost from its initial coating thickness (T0), so that the final resist thickness (T(t)) is given by T(td)=T0−tdR0 (1−exp {−CkP exp [−r2/SS2})q; From this expression one can see how optical exposure (P), development (td, R0) and photoresist thickness (T0) determines final surface relief pattern.
In some aspects, these expose/development processes may be compared with conventional photography. In photography, either exposure or development may be controlled/adjusted as necessary to obtain desired final development pattern. In this sense, one may consider the expose/development level as one process variable which may alternatively be controlled by recording power, development time, developer concentration, etc.
In the mastering process, it is desirable to simultaneously obtain wide lands (for user recorded features) and grooves of suitable depth for adequate tracking signals (e.g., greater than 50 nm). Higher density data storage disks often require the storage or a greater amount of information within the same or smaller size of disk area, resulting in smaller track pitch (i.e., distance between tracks) design criteria.
Attempts have been made to meet these design criteria. In prior art FIGS. 1-3, surface relief patterns of exemplary master disks formed using conventional disk mastering techniques are illustrated using the above expressions to model the effects of exposures from several adjacent tracks recorded in the photoresist layer and then developed. These comparisons assume (1) typical photoresist and developer parameters, (2) constant development time (=40 sec.), (3) SS=0.23 microns, (4) track pitch of 0.375 microns and initial photoresist thickness of 100 nm. As recording power (or alternatively, development time) is increased to obtain deeper grooves, the residual land width diminishes and lands become more rounded due to overlap exposure from adjacent tracks. Partially developed photosensitive material exhibits a granular roughness greater than that of the photosensitive material as initially coated on the disk. Roughness of lands worsens with deepening of grooves, resulting in additional noise in data readback.
More problems occur when the track pitch approaches the finite size of the mastering spot size. For formats where the desired track pitch is much larger (>2×) than the finite size of the mastering spot size (ss), the photosensitive material erosion of the lands is negligible and conventional mastering can provide wide lands with a >50 nm groove depth, or both (due to overlap exposure from adjacent tracks).
In FIG. 4, exemplary embodiments of the mandatory link between land width and groove depth when using conventional mastering processes is illustrated. (Examples of 0.375 micron and 0.425 micron track pitch with 0.22 micron recording spot size). As the groove depth increases, the land width decreases. The master surface relief pattern geometrics (i.e., land width/groove depth) are constrained for given conditions of track pitch and mastering spot size. This means the designer may not independently specify the desired parameters for replica land width and replica groove depth.
A secondary problem for conventional mastering is that the land width precision is limited by mechanical track pitch precision (e.g., mechanical precision of master recording system), which is increasingly difficult to control as track pitch decreases.